#!/usr/bin/gawk -f

BEGIN  { 
	TrainCenter2Inter="trainCenter2Inter.arff";
	TrainCenter2Intra="trainCenter2Intra.arff";
	TestCenter2="testCenter2.arff"; 
	TrainCenter5Inter="trainCenter5Inter.arff";
	TrainCenter5Intra="trainCenter5Intra.arff";
	TestCenter5="testCenter5.arff"; 
	TestSize = 5;
	Seed = 1;
} 

/^[ \t]*$/ {
	next;
} 

NR==1{
	srand(Seed);
	print "" > TrainCenter2Inter;
	print "" > TrainCenter2Intra;
	print "" > TestCenter2;
	print "" > TrainCenter5Inter;
	print "" > TrainCenter5Intra;
	print "" > TestCenter5;
}

FILENAME==ARGV[3] && /@relation/,/@data/ { 
	print $0 >> TrainCenter2Inter;
	print $0 >> TrainCenter2Intra;
	print $0 >> TestCenter2;
	print $0 >> TrainCenter5Inter;
	print $0 >> TrainCenter5Intra;
	print $0 >> TestCenter5;
	next; 
}

FILENAME==ARGV[4] && /@relation/,/@data/ { 
	next; 
}

FILENAME==ARGV[3] {
	InstanceCenter2[rand()] = $0;
}

FILENAME==ARGV[4] {
	InstanceCenter5[rand()] = $0;
}

END {
	TestSizeCenter2 = TestSize;
	for (i in InstanceCenter2)
	{
		if (TestSizeCenter2 > 0)
		{
			print InstanceCenter2[i] >> TestCenter2;
			TestSizeCenter2--;
		}
		else
			print InstanceCenter2[i] >> TrainCenter2Intra;
	}

	for (i in InstanceCenter5)
		print InstanceCenter5[i] >> TrainCenter2Inter;

	TestSizeCenter5 = TestSize;
	for (i in InstanceCenter5)
	{
		if (TestSizeCenter5 > 0)
		{
			print InstanceCenter5[i] >> TestCenter5;
			TestSizeCenter5--;
		}
		else
			print InstanceCenter5[i] >> TrainCenter5Intra;
	}

	for (i in InstanceCenter2)
		print InstanceCenter2[i] >> TrainCenter5Inter;
}